Wind power plant flow field simulation and wind resource evaluation method based on medium and micro-scale coupling

Abstract

The invention discloses a wind power plant flow field simulation and wind resource evaluation method based on mid-scale and micro-scale coupling, and belongs to the technical field of wind power plants, and the method comprises the steps: taking a simulation result of a mid-scale WRF model as a boundary condition of a micro-scale CFD model, simulating the characteristics of a wind power plant flow field, capturing a terrain effect, a fan wake flow and a superimposed effect thereof, and evaluating the wind power plant flow field. Acquiring flow field data of the wind power plant based on medium and micro-scale coupling; performing preprocessing and feature extraction on the wind power plant flow field data to generate wind power plant flow field feature data; and analyzing the wind power plant flow field characteristic data based on the wind resource assessment generating capacity prediction model, and determining a wind resource assessment generating capacity prediction result. According to the method, the problem that the wind power plant planning and optimization effect is poor due to the fact that wind power plant flow field simulation and wind resource effective evaluation cannot be carried out based on medium and micro-scale coupling in the prior art is solved. Wind power plant flow field simulation and wind resource effective evaluation are carried out based on medium and micro-scale coupling, and the wind power plant planning and optimization effect can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wind farms, in particular to a wind farm flow field simulation and wind resource assessment method based on meso-micro scale coupling. BACKGROUND

[0002] A wind farm is a place where wind power is concentratedly constructed and utilized, which is of great significance for the utilization of renewable energy and environmental protection. The wind farm flow field simulation and wind resource assessment are important links in the planning and optimization of wind farms. Traditional methods usually use single-scale models, such as mesoscale or microscale.

[0003] The microscale CFD model can finely analyze the flow changes caused by topography, buildings and wind turbine wake, but its boundary conditions are usually assumed to be uniform and stable inflow wind, which cannot reflect the atmospheric boundary layer characteristics that change over time in the real atmosphere. The mesoscale WRF model can simulate large-scale weather processes and regional climate, but its grid is relatively coarse and cannot accurately analyze the flow details of complex terrain and wind turbine scales. The ability to simulate wind turbine wake is very limited. Therefore, it is particularly important to simulate the wind farm flow field and assess the wind resources based on meso-micro scale coupling.

[0004] The existing technology cannot effectively simulate the wind farm flow field and assess the wind resources based on meso-micro scale coupling, resulting in poor planning and optimization of wind farms. SUMMARY

[0005] The present application aims to provide a wind farm flow field simulation and wind resource assessment method based on meso-micro scale coupling, which can effectively simulate the wind farm flow field and assess the wind resources based on meso-micro scale coupling, and improve the planning and optimization of wind farms, solving the problems raised in the background technology.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0007] The wind farm flow field simulation and wind resource assessment method based on meso-micro scale coupling comprises:

[0008] The simulation results of the mesoscale WRF model are used as the boundary conditions of the microscale CFD model based on meso-micro scale coupling. By simulating the flow field characteristics of the wind farm, the topographic effect, wind turbine wake and their superposition effect are captured, and the wind farm flow field data based on meso-micro scale coupling are obtained.

[0009] The wind farm flow field data based on meso-micro scale coupling are preprocessed and feature extracted based on natural language processing technology, and wind farm flow field feature data are generated.

[0010] The wind resource assessment power generation prediction model is constructed, the wind resource assessment power generation prediction result is determined by analyzing the flow field characteristic data of the wind farm based on the wind resource assessment power generation prediction model, and the wind resource assessment of the wind farm region is performed.

[0011] Preferably, based on the meso-micro scale coupling, the simulation result of the meso-scale WRF model is taken as the boundary condition of the micro-scale CFD model, the flow field characteristics of the wind farm are simulated, the terrain effect, the wind turbine wake and the superposition effect thereof are captured, the wind farm flow field data based on the meso-micro scale coupling are obtained, and the following operations are performed:

[0012] The atmospheric circulation of the wind farm region is simulated based on the meso-scale WRF model, and long-time sequence meteorological data of the wind farm region are extracted, including wind speed, wind direction, temperature, air pressure and turbulence intensity;

[0013] The extracted meteorological data is taken as the boundary condition of the micro-scale CFD model, a three-dimensional linear interpolation method is used to interpolate the data of the meso-scale WRF model from its own grid and coordinate system to the grid and coordinate system of the micro-scale CFD model, so that the data of the meso-scale WRF model is mapped to the boundary of the micro-scale CFD model;

[0014] The flow field characteristics around the wind turbine of the wind farm region are simulated based on the micro-scale CFD model, the terrain effect, the wind turbine wake and the superposition effect thereof are captured, and the wind farm flow field data based on the meso-micro scale coupling are obtained.

[0015] Preferably, based on the micro-scale CFD model, the flow field characteristics around the wind turbine of the wind farm region are simulated, and the following operations are performed:

[0016] An initial micro-scale WRF model is determined based on the area and height of the wind farm region where the target wind farm is located;

[0017] Terrain information of the wind farm region where the target wind farm is located is obtained, and the type of the electric field terrain of the wind farm region where the target wind farm is located is determined based on the terrain information;

[0018] The complexity of the terrain of the electric field terrain type is determined, and the first model parameter of the initial micro-scale WRF model of the target wind farm is adjusted according to the complexity of the terrain, to obtain a first micro-scale WRF model, wherein the first model parameter is used to represent the grid resolution of the initial micro-scale WRF model;

[0019] The second model parameter of the first micro-scale WRF model is adjusted based on the land surface hydrology and thermal information of the wind farm region where the target wind farm is located, to obtain a second micro-scale WRF model, wherein the second model parameter is used to represent the model accuracy of the near-ground region of the first meso-micro scale WRF model;

[0020] Obtaining the running state of the wind turbine of the target wind farm in a preset long time sequence, so as to obtain the airflow fluctuation of the wind farm region where the target wind farm is located under different wind turbine running states;

[0021] According to the influence degree of the airflow fluctuation under different wind turbine running states on the flow field in the wind farm region, the second micro-scale WRF model under different wind turbine running states is optimized respectively, and a micro-scale WRF model set of the target wind farm region is obtained;

[0022] Based on the running state of the wind turbine at each monitoring moment, the micro-scale WRF model corresponding to the state is selected from the micro-scale WRF model set;

[0023] If the wind turbine of the target wind farm is in working state, the wind turbine running parameter of the target wind farm is obtained, and the flow field characteristics around the wind turbine in the wind farm region at the corresponding monitoring moment are simulated based on the wind turbine running parameter and the micro-scale WRF model corresponding to the state;

[0024] On the contrary, the flow field characteristics around the wind turbine in the wind farm region at the corresponding monitoring moment are simulated based on the micro-scale WRF model corresponding to the state.

[0025] Preferably, the wind farm flow field data based on meso-micro scale coupling is preprocessed based on natural language processing technology, and the following operations are performed:

[0026] The wind farm flow field data based on meso-micro scale coupling is cleaned to remove noise in the wind farm flow field data based on meso-micro scale coupling and reduce the interference of noise data on wind farm wind resource assessment;

[0027] The wind farm flow field data based on meso-micro scale coupling is checked to identify abnormal values in the wind farm flow field data based on meso-micro scale coupling, and the abnormal values in the wind farm flow field data are processed;

[0028] Among them, the abnormal values in the wind farm flow field data are evaluated based on the abnormal type to determine whether the abnormal values in the wind farm flow field data are valuable for wind farm wind resource assessment;

[0029] If the abnormal values in the wind farm flow field data are valuable for wind farm wind resource assessment, the abnormal values in the wind farm flow field data are replaced, otherwise the abnormal values in the wind farm flow field data are deleted.

[0030] Preferably, the wind farm flow field data based on meso-micro scale coupling is extracted based on natural language processing technology, and the following operations are performed:

[0031] The wind farm flow field data based on the meso-micro scale coupling is standardized, so that the wind farm flow field data based on the meso-micro scale coupling is converted into a unified data format, the dimensional differences between the wind farm flow field data are eliminated, and the standardized wind farm flow field data are formed.

[0032] The standardized wind farm flow field data are subjected to feature extraction, so that the feature vectors related to the wind farm wind resource assessment are extracted from the standardized wind farm flow field data, and the wind farm flow field feature data are determined.

[0033] Preferably, a wind resource assessment power generation prediction model is constructed, and the following operations are performed:

[0034] Wind resource assessment historical data of a wind farm are collected, and the collected wind resource assessment historical data of the wind farm are divided, wherein the wind resource assessment historical data of the wind farm are divided into a training set and a test set;

[0035] The training set is used to train a deep learning model, so that the deep learning model autonomously learns the wind resource assessment power generation prediction behavior from the training set, automatically assesses the wind resource of the wind farm and predicts the power generation of the wind farm, and a wind resource assessment power generation prediction model is determined;

[0036] The test set is used to test the wind resource assessment power generation prediction model, and the generalization performance of the wind resource assessment power generation prediction model is evaluated, and a model test evaluation result is determined;

[0037] According to the model test evaluation result, the wind resource assessment power generation prediction model is optimized, and the optimal wind resource assessment power generation prediction model is determined.

[0038] Preferably, the generalization performance of the wind resource assessment power generation prediction model is evaluated, and the following operations are performed:

[0039] The test set is input into the wind resource assessment power generation prediction model, the generalization performance of the wind resource assessment power generation prediction model is evaluated based on the accuracy, precision or recall rate, and it is judged whether the wind resource assessment power generation prediction model can automatically assess the wind resource of the wind farm and predict the power generation of the wind farm;

[0040] When the wind resource assessment power generation prediction model cannot automatically assess the wind resource of the wind farm and predict the power generation of the wind farm, the wind resource assessment power generation prediction model is adjusted and optimized in parameters, and it is judged whether the optimized wind resource assessment power generation prediction model can automatically assess the wind resource of the wind farm and predict the power generation of the wind farm, until the wind resource assessment power generation prediction model can automatically assess the wind resource of the wind farm and predict the power generation of the wind farm, and the optimal wind resource assessment power generation prediction model is determined.

[0041] Preferably, the generalization performance of the wind resource assessment power generation prediction model is evaluated based on the accuracy, precision or recall rate, and the following operations are performed:

[0042] Each set of input test data in the test set is input into the power generation prediction model respectively, and the difference between the power generation prediction result of each set of input test data and the corresponding output test data is compared, so as to obtain the accuracy, precision or recall rate of the power generation prediction model;

[0043] The comprehensive generalization index of the power generation prediction model is determined based on the accuracy, precision or recall rate of the power generation prediction model, and the generalization performance of the power generation prediction model is judged based on the comprehensive generalization index.

[0044] Preferably, the wind resource assessment power generation prediction model is used to analyze the wind farm flow field characteristic data, and the following operations are performed:

[0045] The wind farm flow field characteristic data is input into the wind resource assessment power generation prediction model, the wind farm flow field characteristic data is analyzed according to the wind resource assessment power generation prediction model, the wind resource of the wind farm is automatically evaluated and the power generation of the wind farm is predicted, and the wind resource assessment power generation prediction result is determined.

[0046] Preferably, according to the wind resource assessment power generation prediction result and in combination with the wind farm flow field characteristic data, a wind resource assessment report of the wind farm is generated, the wind resource in the wind farm area is evaluated, and the wind resource assessment report of the wind farm is displayed in a visual form, thereby providing strong support for the site selection of the wind farm.

[0047] Compared with the prior art, the present application has the following advantages:

[0048] The present application simulates the atmospheric circulation in the wind farm area by using the mesoscale WRF model, extracts long-time sequence meteorological data of the wind farm area, uses the extracted meteorological data as the boundary condition of the microscale CFD model, uses the three-dimensional linear interpolation method to map the data of the mesoscale WRF model to the boundary of the microscale CFD model, simulates the flow field characteristics around the wind turbine in the wind farm area based on the microscale CFD model, captures the terrain effect, the wake of the wind turbine and the superposition effect thereof, obtains the wind farm flow field data based on the meso-microscale coupling, pre-processes and extracts features from the wind farm flow field data based on the meso-microscale coupling based on the natural language processing technology, generates the wind farm flow field characteristic data, analyzes the wind farm flow field characteristic data based on the wind resource assessment power generation prediction model, determines the wind resource assessment power generation prediction result, evaluates the wind resource in the wind farm area, and provides strong support for the site selection of the wind farm. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 A flowchart of a method for wind farm flow field simulation and wind resource assessment based on meso-micro scale coupling of the present application. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0051] To solve the problem that the existing wind farm flow field simulation and wind resource effective assessment cannot be based on meso-micro scale coupling, resulting in poor wind farm planning and optimization effect, please refer to Figure 1 The technical solutions provided in the embodiments are as follows:

[0052] The method for wind farm flow field simulation and wind resource assessment based on meso-micro scale coupling comprises the following steps.

[0053] Based on meso-micro scale coupling, the simulation results of the mesoscale WRF model are taken as the boundary conditions of the microscale CFD model, the wind farm flow field characteristics are simulated, the terrain effect, wind turbine wake and their superposition effect are captured, and the wind farm flow field data based on meso-micro scale coupling are obtained.

[0054] The mesoscale WRF model is used to simulate macro atmospheric circulation, including weather system background information such as wind speed, wind direction and pressure distribution, and provides overall wind field characteristics of the area where the wind farm is located as the input boundary conditions of the microscale CFD model.

[0055] The microscale CFD model is used to simulate the flow field characteristics around the wind turbine, including turbulence, wake effect and aerodynamic load of the wind turbine blade, and analyze the wind turbine layout, wake interference and local wind speed distribution.

[0056] By combining the mesoscale WRF model and the microscale CFD model, a connection between macro atmospheric circulation and micro wind field characteristics can be established, so that higher precision simulation and evaluation can be realized. The mesoscale WRF model provides a real, dynamic and time-varying atmospheric background field, and the microscale CFD model performs high-resolution simulation on the wind farm area under the refined boundary conditions provided by the mesoscale WRF model, accurately capturing the terrain effect, wind turbine wake and their superposition effect.

[0057] In the embodiments, based on meso-micro scale coupling, the simulation results of the mesoscale WRF model are taken as the boundary conditions of the microscale CFD model, the wind farm flow field characteristics are simulated, the terrain effect, wind turbine wake and their superposition effect are captured, and the wind farm flow field data based on meso-micro scale coupling are obtained, and the following operations are performed:

[0058] Based on the mesoscale WRF model, the atmospheric circulation of the wind farm area is simulated, and the long-time sequence meteorological data of the wind farm area is extracted, including wind speed, wind direction, temperature, air pressure and turbulence intensity;

[0059] The extracted meteorological data is used as the boundary condition of the microscale CFD model, and the three-dimensional linear interpolation method is used to interpolate the data of the mesoscale WRF model from its own grid and coordinate system to the grid and coordinate system of the microscale CFD model, so that the data of the mesoscale WRF model is mapped to the boundary of the microscale CFD model;

[0060] Based on the microscale CFD model, the flow field characteristics around the wind turbine in the wind farm area are simulated, the terrain effect, wind turbine wake and superposition effect are captured, and the wind farm flow field data based on mesoscale-microscale coupling are obtained.

[0061] The terrain effect is that the model automatically simulates the acceleration of airflow on an uphill, the deceleration on a downhill, the separation at the ridge and the channel effect in the valley.

[0062] The wind turbine wake is that the model shows the velocity deficit zone (wake) formed behind each wind turbine, the enhanced turbulence intensity zone and the deflection of the wake.

[0063] The superposition effect is: 1) Wake-wake interaction: When the downstream wind turbine is located in the wake of the upstream wind turbine, the inflow wind speed is lower and the turbulence intensity is higher, and the microscale CFD model can accurately capture this complex interaction, including the merging and deflection of the wake; 2) Wake-terrain interaction: When the wake passes through a complex terrain, it may be distorted, accelerated or lifted, thereby changing its influence on the downstream wind turbine, and the microscale CFD model can capture this phenomenon.

[0064] Based on natural language processing technology, the wind farm flow field data based on mesoscale-microscale coupling are preprocessed and feature extracted to generate wind farm flow field feature data.

[0065] In this embodiment, based on the microscale CFD model, the flow field characteristics around the wind turbine in the wind farm area are simulated, and the following operations are also performed:

[0066] An initial microscale WRF model is determined based on the area and height of the wind farm area where the target wind farm is located;

[0067] Obtain the terrain information of the wind farm area where the target wind farm is located, and determine the wind farm terrain type of the wind farm area where the target wind farm is located based on the terrain information;

[0068] determine a terrain complexity of the electric field terrain type, and adjust a first model parameter of an initial micro-scale WRF model of the target wind farm according to the terrain complexity, to obtain a first micro-scale WRF model, wherein the first model parameter is used to represent a grid resolution of the initial micro-scale WRF model;

[0069] adjust a second model parameter of the first micro-scale WRF model based on land surface hydrology and thermal information of a wind farm region where the target wind farm is located, to obtain a second micro-scale WRF model, wherein the second model parameter is used to represent a model accuracy of a near-ground region of the first micro-scale WRF model;

[0070] obtain a wind turbine running state of the target wind farm in a preset long time sequence, to obtain airflow fluctuation of the wind farm region where the target wind farm is located under different wind turbine running states;

[0071] respectively optimize the second micro-scale WRF model under different wind turbine running states according to an influence degree of the airflow fluctuation under the different wind turbine running states on a flow field in the wind farm region, to obtain a micro-scale WRF model set of the target wind farm region;

[0072] select a micro-scale WRF model corresponding to a state from the micro-scale WRF model set based on a wind turbine running state of each monitoring moment;

[0073] if the wind turbine of the target wind farm is in a working state, obtain a wind turbine running parameter of the target wind farm, and simulate a flow field characteristic around a wind turbine of the wind farm region at a corresponding monitoring moment based on the wind turbine running parameter and the micro-scale WRF model corresponding to the state;

[0074] otherwise, simulate the flow field characteristic around the wind turbine of the wind farm region at the corresponding monitoring moment based on the micro-scale WRF model corresponding to the state.

[0075] In this embodiment, the initial micro-scale WRF model is a basic numerical model constructed based on a spatial range of the target wind farm region, and is used to simulate a flow field characteristic in an atmospheric boundary layer. Core parameters thereof include a grid resolution, a vertical layer number, a time step, and the like, wherein the spatial range includes a horizontal area and a vertical height. For example, the vertical layer of the initial micro-scale WRF model can be a bottom layer of 5m resolution (0-100m), an upper layer logarithmically stretched to 2km, and a total of 60 layers.

[0076] In this embodiment, the electric field terrain type is a category divided according to terrain features (such as an altitude, a slope, and a surface coverage) of the wind farm region, and directly affects flow field parameters such as a wind speed and a turbulence intensity. For example, the electric field terrain type can be divided into a flat terrain, a complex terrain, and a near-sea terrain, and the complex terrain can be a terrain with a slope greater than 15%, including a mountain, a canyon, or a forest.

[0077] In this embodiment, the terrain complexity degree is used to quantify the intensity of the terrain interference to the airflow, which is usually evaluated by indicators such as slope, elevation variability, and surface roughness. Complex terrain can enhance turbulence and change wind direction, and the grid resolution of the model needs to be adjusted to capture the details. For example, the grid resolution of the model can be set to 200 m for a flat terrain with low complexity, and the grid resolution needs to be increased to 50 m for a complex terrain with a higher terrain complexity degree.

[0078] In this embodiment, the first model parameter is a parameter for controlling the spatial discretization degree of the model, which directly affects the calculation accuracy and efficiency of the microscale model. For example, the grid in the initial microscale WRF model is 100 m, and the grid in the first microscale WRF model is 50 m.

[0079] In this embodiment, the land surface hydrology and thermal information is a parameter for reflecting the energy exchange between the ground surface and the atmosphere, including soil moisture, vegetation coverage, and ground surface temperature. The land surface hydrology and thermal information affects the near-surface wind speed and turbulence through the land surface model (such as Noah-MP). For example, in arid regions, the land surface hydrology and thermal information has low soil moisture, small surface flux, and weak wind speed attenuation.

[0080] In this embodiment, the second model parameter is a parameter for controlling the simulation precision of the near-surface layer (0-100 m), including the turbulence closure scheme and the surface flux calculation method. High precision requires the use of LES mode or dynamic subgrid model.

[0081] In this embodiment, the wind turbine operating state includes start-up, shutdown, and variable pitch, etc. The wind turbine operating state is determined based on the wind turbine power curve, pitch angle, and wind measurement data, etc.

[0082] In this embodiment, the wind turbine operating state is different, and the influence on the downstream wake is different. For example, in the start-up state of the wind turbine, the wind speed in the wake area decreases by 20%, and the turbulence intensity increases by 50%.

[0083] In this embodiment, the microscale WRF model set is a plurality of models optimized for different wind turbine operating states, and each model corresponds to a specific state parameterization scheme of the flow field. For example, in the shutdown state of the wind turbine, the ALM is closed, and only the natural wind field is simulated; in the start-up state of the wind turbine, the actuator line model is enabled; in the variable pitch adjustment state, the thrust coefficient is dynamically adjusted, etc.

[0084] In this embodiment, the model screening is to select a matching microscale WRF model from the microscale WRF model set for flow field simulation according to the real-time monitored wind turbine operating state. For example, the wind turbine of the target wind farm is shut down due to high wind speed, and the "shutdown state model" in the microscale WRF model set is selected.

[0085] In this embodiment, the fan operating parameters are fan parameters that affect the fan wake, including tip speed ratio, pitch angle, and rotating speed, etc.

[0086] The working principle of the above technical solution is as follows: first, an initial micro-scale WRF model is constructed based on the horizontal and vertical spatial range of the wind farm, and the grid resolution of the model is adjusted in combination with the terrain classification result of the wind farm area; second, a land surface model is introduced to determine the land surface hydrological thermal information, optimize the turbulence closure scheme of the near-surface area, and improve the model accuracy of the micro-scale WRF model of the near-surface area; third, the running state of the fan, such as start-stop and variable pitch, is judged, so as to construct a micro-scale WRF model set under different fan running states according to the influence of different fan running states on the wake; when real-time simulation is performed based on the micro-scale WRF model, the corresponding micro-scale WRF model is automatically selected according to the monitored fan running state, if the fan is running, the wake field is corrected by superimposing the fan dynamic parameters such as tip speed ratio and thrust coefficient, otherwise the conventional micro-scale WRF model is directly called, thus realizing millimeter-level dynamic simulation of the flow field of the wind farm in a complex environment.

[0087] The beneficial effects of the above technical solution are as follows: through adaptive grid densification according to the complexity of the terrain, dynamic correction of the micro-scale WRF model by the land surface thermal parameters, and model switching in the micro-scale WRF model set driven by different fan states, the flow field characteristic simulation accuracy of the micro-scale WRF model is significantly improved, thus providing more reliable data support for wind farm power prediction, wake control, and wind farm fan layout optimization.

[0088] In this embodiment, the natural language processing technology is used to preprocess the wind farm flow field data based on meso-micro scale coupling, and the following operations are performed:

[0089] The wind farm flow field data based on meso-micro scale coupling is cleaned to remove noise in the wind farm flow field data based on meso-micro scale coupling and reduce the interference of noise data on wind farm wind resource assessment;

[0090] The wind farm flow field data based on meso-micro scale coupling is checked to identify abnormal values in the wind farm flow field data based on meso-micro scale coupling, and the abnormal values in the wind farm flow field data are processed;

[0091] The abnormal values in the wind farm flow field data are evaluated based on the abnormal type to determine whether the abnormal values in the wind farm flow field data are valuable for wind farm wind resource assessment;

[0092] If the abnormal values in the wind farm flow field data are valuable for wind farm wind resource assessment, the abnormal values in the wind farm flow field data are replaced, otherwise the abnormal values in the wind farm flow field data are deleted.

[0093] Specifically, by cleaning the wind farm flow field data based on meso-micro scale coupling, the noise in the wind farm flow field data can be removed, and the abnormal values in the wind farm flow field data can be processed, so as to improve the data quality of the wind farm flow field data.

[0094] In this embodiment, based on the natural language processing technology, the wind farm flow field data based on meso-micro scale coupling is extracted, and the following operations are performed:

[0095] The wind farm flow field data based on meso-micro scale coupling is standardized, so that the wind farm flow field data based on meso-micro scale coupling is converted into a unified data format, the dimensional difference between the wind farm flow field data is eliminated, and standardized wind farm flow field data is formed.

[0096] The standardized wind farm flow field data is extracted, and the feature vector related to the wind resource assessment of the wind farm is extracted from the standardized wind farm flow field data, and the wind farm flow field feature data is determined.

[0097] Specifically, by standardizing and extracting the wind farm flow field data based on meso-micro scale coupling, wind farm flow field feature data can be generated, which facilitates better assessment of wind resources and prediction of wind farm power generation in the subsequent process.

[0098] A wind resource assessment and power generation prediction model is constructed, the wind farm flow field feature data is analyzed based on the wind resource assessment and power generation prediction model, the wind resource assessment and power generation prediction result is determined, and the wind resource assessment in the wind farm area is performed.

[0099] In this embodiment, the wind resource assessment and power generation prediction model is constructed, and the following operations are performed:

[0100] The wind farm wind resource assessment historical data is collected, and the collected wind farm wind resource assessment historical data is divided, wherein the wind farm wind resource assessment historical data is divided into a training set and a test set;

[0101] The training set is used to train the deep learning model, so that the deep learning model learns the wind resource assessment and power generation prediction behavior from the training set, automatically assesses the wind resources of the wind farm and predicts the power generation of the wind farm, and determines the wind resource assessment and power generation prediction model;

[0102] The test set is used to test the wind resource assessment and power generation prediction model, and the generalization performance of the wind resource assessment and power generation prediction model is evaluated, and the model test evaluation result is determined;

[0103] According to the model test evaluation result, the wind resource assessment and power generation prediction model is optimized, and the optimal wind resource assessment and power generation prediction model is determined.

[0104] In the embodiment, the generalization performance of the wind resource assessment power generation prediction model is evaluated, and the following operations are performed:

[0105] The test set is input into the wind resource assessment power generation prediction model, the generalization performance of the wind resource assessment power generation prediction model is evaluated based on the accuracy, precision or recall, and it is determined whether the wind resource assessment power generation prediction model can automatically assess the wind resource of the wind farm and predict the power generation of the wind farm;

[0106] When the wind resource assessment power generation prediction model cannot automatically assess the wind resource of the wind farm and predict the power generation of the wind farm, the wind resource assessment power generation prediction model is adjusted and optimized, and it is determined whether the optimized wind resource assessment power generation prediction model can automatically assess the wind resource of the wind farm and predict the power generation of the wind farm, until the wind resource assessment power generation prediction model can automatically assess the wind resource of the wind farm and predict the power generation of the wind farm, and the optimal wind resource assessment power generation prediction model is determined.

[0107] In the embodiment, the generalization performance of the wind resource assessment power generation prediction model is evaluated based on the accuracy, precision or recall, and the following operations are performed:

[0108] Each group of input test data in the test set is input into the power generation prediction model, and the data difference between the power generation prediction result of each group of input test data and the corresponding output test data is compared, so as to obtain the accuracy, precision or recall of the power generation prediction model;

[0109] The comprehensive generalization index of the power generation prediction model is determined based on the accuracy, precision or recall of the power generation prediction model, and the generalization performance of the power generation prediction model is determined based on the comprehensive generalization index.

[0110] The calculation process of the above-mentioned comprehensive generalization index is as follows:

[0111]

[0112] wherein, is the comprehensive generalization index of the power generation prediction model, is the amount of test data in which the power generation prediction model correctly predicts the power generation of the test data to meet the standard, is the amount of test data in which the power generation prediction model correctly predicts the power generation of the test data to not meet the standard, is the amount of test data in which the power generation prediction model incorrectly predicts the power generation of the test data to meet the standard, is the amount of test data in which the power generation prediction model incorrectly predicts the power generation of the test data to not meet the standard, is the accuracy of the power generation prediction model, is the precision of the power generation prediction model, is the recall of the power generation prediction model, and the data amount of the test data is , a first influence weight corresponding to the precision rate, a second influence weight corresponding to the precision rate, a third influence weight corresponding to the recall rate, wherein the sum of the weights of the first influence weight, the second influence weight and the third influence weight is 1, and any one of the first influence weight, the second influence weight and the third influence weight can be 0.

[0113] According to the above technical means, the application can obtain the accuracy rate, the precision rate or the recall rate of the power generation prediction model by inputting the test data into the power generation prediction model, so as to determine the comprehensive generalization index of the power generation prediction model, quantify the generalization performance of the power generation prediction model, and more effectively judge the model performance of the power generation prediction model, so that the prediction of the power generation of the wind farm is more accurate and effective.

[0114] In the embodiment, the wind resource assessment power generation prediction model is used to analyze the flow field characteristic data of the wind farm, and the following operations are performed:

[0115] The flow field characteristic data of the wind farm is input into the wind resource assessment power generation prediction model, the flow field characteristic data of the wind farm is analyzed according to the wind resource assessment power generation prediction model, the wind resource of the wind farm is automatically assessed and the power generation of the wind farm is predicted, and the wind resource assessment power generation prediction result is determined.

[0116] In the embodiment, the wind resource assessment power generation prediction result is combined with the flow field characteristic data of the wind farm to generate a wind resource assessment report of the wind farm, the wind resource in the wind farm area is assessed, and the wind resource assessment report of the wind farm is displayed in a visual form, which provides strong support for the site selection of the wind farm.

[0117] In summary, the atmospheric circulation in the wind farm area is simulated by the mesoscale WRF model, the long-time sequence meteorological data of the wind farm area is extracted, the extracted meteorological data is used as the boundary condition of the microscale CFD model, the three-dimensional linear interpolation method is used to map the data of the mesoscale WRF model to the boundary of the microscale CFD model, the flow field characteristics around the wind turbine in the wind farm area are simulated based on the microscale CFD model, the terrain effect, the wind turbine wake and their superposition effect are captured, the wind farm flow field data based on the meso-microscale coupling is obtained, the wind farm flow field data based on the meso-microscale coupling is preprocessed and feature extracted based on the natural language processing technology, the wind farm flow field characteristic data is generated, the wind resource assessment power generation prediction model is used to analyze the wind farm flow field characteristic data, the wind resource assessment power generation prediction result is determined, the wind resource in the wind farm area is assessed, and strong support is provided for the site selection of the wind farm. The wind farm flow field simulation and the effective wind resource assessment can be performed based on the meso-microscale coupling, and the planning and optimization effect of the wind farm can be improved.

[0118] It has to be noted that, in the present document, the terms "first", "second", etc. merely serve to identify a subject or action, without necessarily requiring or implying any such actual relationship or order between such subjects or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0119] While embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, combinations, and variations of the embodiments can be undertaken without departing from the spirit and scope of the present application, which is defined by the appended claims and their equivalents.

Claims

1. A method for wind farm flow field simulation and wind resource assessment based on meso-micro scale coupling, characterized in that, Comprise: Based on the coupling of mesoscale and microscale, the simulation results of the mesoscale WRF model are taken as the boundary conditions of the microscale CFD model, the flow field characteristics of the wind farm are simulated, the terrain effect, the wake of the wind turbine and their superposition effect are captured, and the wind farm flow field data based on the coupling of mesoscale and microscale are obtained; Based on the natural language processing technology, the wind farm flow field data based on the coupling of mesoscale and microscale are preprocessed and feature extracted, and the wind farm flow field feature data are generated; The wind resource assessment power generation prediction model is constructed, the wind farm flow field feature data are analyzed based on the wind resource assessment power generation prediction model, the wind resource assessment power generation prediction result is determined, and the wind resource assessment in the wind farm area is performed.

2. The meso-microscale coupling based wind farm flow field simulation and wind resource assessment method according to claim 1, characterized in that, Based on the coupling of mesoscale and microscale, the simulation results of the mesoscale WRF model are taken as the boundary conditions of the microscale CFD model, the flow field characteristics of the wind farm are simulated, the terrain effect, the wake of the wind turbine and their superposition effect are captured, and the wind farm flow field data based on the coupling of mesoscale and microscale are obtained, and the following operations are performed: The atmospheric circulation of the wind farm area is simulated based on the mesoscale WRF model, and the long time sequence meteorological data of the wind farm area are extracted, including wind speed, wind direction, temperature, air pressure and turbulence intensity; The extracted meteorological data are taken as the boundary conditions of the microscale CFD model, a three-dimensional linear interpolation method is used to interpolate the data of the mesoscale WRF model from its own grid and coordinate system to the grid and coordinate system of the microscale CFD model, so that the data of the mesoscale WRF model are mapped to the boundary of the microscale CFD model; The flow field characteristics around the wind turbine in the wind farm area are simulated based on the microscale CFD model, the terrain effect, the wake of the wind turbine and their superposition effect are captured, and the wind farm flow field data based on the coupling of mesoscale and microscale are obtained.

3. The meso-microscale coupling based wind farm flow field simulation and wind resource assessment method according to claim 2, characterized in that, Based on the microscale CFD model, the flow field characteristics around the wind turbine in the wind farm area are simulated, and the following operations are also performed: Determine the initial microscale WRF model based on the area and height of the wind farm area where the target wind farm is located; Obtain the terrain information of the wind farm area where the target wind farm is located, and determine the electric field terrain type of the wind farm area where the target wind farm is located based on the terrain information; Determine the terrain complexity of the electric field terrain type, and adjust the first model parameter of the initial microscale WRF model of the target wind farm according to the terrain complexity to obtain the first microscale WRF model, wherein the first model parameter is used to represent the grid resolution of the initial microscale WRF model; Adjust the second model parameter of the first microscale WRF model based on the land surface hydrology and thermal information of the wind farm area where the target wind farm is located, to obtain the second microscale WRF model, wherein the second model parameter is used to represent the model accuracy of the near-ground area of the first mesoscale WRF model; Obtain the wind turbine operating state of the target wind farm in a predetermined long time sequence, so as to obtain the airflow fluctuation of the wind farm area where the target wind farm is located under different wind turbine operating states; According to the influence degree of the airflow fluctuation of the wind turbine under different operating states on the flow field in the wind farm area, the second micro-scale WRF model under different operating states of the wind turbine is optimized to obtain a micro-scale WRF model set of the target wind farm area; Based on the operating state of the wind turbine at each monitoring moment, a micro-scale WRF model corresponding to the state is selected from the micro-scale WRF model set; If the wind turbine of the target wind farm is in a working state, the operating parameters of the wind turbine of the target wind farm are obtained, and the flow field characteristics around the wind turbine in the wind farm area at the corresponding monitoring moment are simulated based on the operating parameters of the wind turbine and the micro-scale WRF model corresponding to the state; On the contrary, the flow field characteristics around the wind turbine in the wind farm area at the corresponding monitoring moment are simulated based on the micro-scale WRF model corresponding to the state.

4. The meso-micro coupled wind farm flow field simulation and wind resource assessment method according to claim 2, wherein, The wind farm flow field data based on the meso-micro scale coupling are preprocessed based on the natural language processing technology, and the following operations are performed: The wind farm flow field data based on the meso-micro scale coupling are cleaned to remove noise in the wind farm flow field data based on the meso-micro scale coupling and reduce the interference of noise data on the wind resource assessment of the wind farm; The wind farm flow field data based on the meso-micro scale coupling are checked to identify abnormal values in the wind farm flow field data based on the meso-micro scale coupling, and the abnormal values in the wind farm flow field data are processed; Among them, the abnormal values in the wind farm flow field data are evaluated based on the abnormal type to determine whether the abnormal values in the wind farm flow field data are valuable for the wind resource assessment of the wind farm; If the abnormal values in the wind farm flow field data are valuable for the wind resource assessment of the wind farm, the abnormal values in the wind farm flow field data are replaced, otherwise the abnormal values in the wind farm flow field data are deleted.

5. The meso-microscale coupling based wind farm flow field simulation and wind resource assessment method according to claim 4, characterized in that, The wind farm flow field data based on the meso-micro scale coupling are preprocessed based on the natural language processing technology, and the following operations are performed: The wind farm flow field data based on the meso-micro scale coupling are standardized to convert the wind farm flow field data based on the meso-micro scale coupling into a unified data format, eliminate the dimension difference between the wind farm flow field data, and form standardized wind farm flow field data; The standardized wind farm flow field data are feature extracted to extract feature vectors related to the wind resource assessment of the wind farm from the standardized wind farm flow field data, and determine the wind farm flow field feature data.

6. The meso-microscale coupling based wind farm flow field simulation and wind resource assessment method according to claim 5, characterized in that, A wind resource assessment power generation prediction model is constructed, and the following operations are performed: Collect wind resource assessment historical data of the wind farm, and divide the collected wind resource assessment historical data of the wind farm, wherein the wind resource assessment historical data of the wind farm is divided into a training set and a test set; The training set is used to train the deep learning model, so that the deep learning model learns the wind resource assessment power generation prediction behavior from the training set, automatically assesses the wind resource of the wind farm, and predicts the power generation of the wind farm, and determines the wind resource assessment power generation prediction model; The test set is used to test the wind resource assessment power generation prediction model, and the generalization performance of the wind resource assessment power generation prediction model is evaluated to determine the model test evaluation result; According to the model test evaluation result, the wind resource assessment power generation prediction model is optimized, and the optimal wind resource assessment power generation prediction model is determined.

7. The meso-microscale coupling based wind farm flow field simulation and wind resource assessment method according to claim 6, characterized in that, The generalization performance of the wind resource assessment power generation prediction model is evaluated, and the following operations are performed: The test set is input into the wind resource assessment power generation prediction model, and the generalization performance of the wind resource assessment power generation prediction model is evaluated based on the accuracy, precision or recall rate, and it is judged whether the wind resource assessment power generation prediction model can automatically assess the wind resource of the wind farm and predict the power generation of the wind farm; When the wind resource assessment power generation prediction model cannot automatically assess the wind resource of the wind farm and predict the power generation of the wind farm, the parameters of the wind resource assessment power generation prediction model are adjusted and optimized, and it is judged whether the optimized wind resource assessment power generation prediction model can automatically assess the wind resource of the wind farm and predict the power generation of the wind farm, until the wind resource assessment power generation prediction model can automatically assess the wind resource of the wind farm and predict the power generation of the wind farm, and the optimal wind resource assessment power generation prediction model is determined.

8. The meso-microscale coupling based wind farm flow field simulation and wind resource assessment method according to claim 7, characterized in that, The generalization performance of the wind resource assessment power generation prediction model is evaluated based on the accuracy, precision or recall rate, and the following operations are performed: Each group of input test data in the test set is input into the power generation prediction model, and the data difference between the power generation prediction result of each group of input test data and the corresponding output test data is compared, so as to obtain the accuracy, precision or recall rate of the power generation prediction model; Based on the accuracy, precision or recall rate of the power generation prediction model, the comprehensive generalization index of the power generation prediction model is determined, and the generalization performance of the power generation prediction model is judged based on the comprehensive generalization index.

9. The meso-microscale coupling based wind farm flow field simulation and wind resource assessment method according to claim 7, characterized in that, Based on the wind resource assessment power generation prediction model, the flow field characteristic data of the wind farm is analyzed, and the following operations are performed: The flow field characteristic data of the wind farm is input into the wind resource assessment power generation prediction model, the flow field characteristic data of the wind farm is analyzed according to the wind resource assessment power generation prediction model, the wind resource of the wind farm is automatically assessed and the power generation of the wind farm is predicted, and the wind resource assessment power generation prediction result is determined.

10. The meso-microscale coupling based wind farm flow field simulation and wind resource assessment method according to claim 9, characterized in that, According to the wind resource assessment power generation prediction result and combined with the flow field characteristic data of the wind farm, a wind resource assessment report of the wind farm is generated, the wind resource in the wind farm area is assessed, and the wind resource assessment report of the wind farm is displayed in a visual form, which provides strong support for the site selection of the wind farm.

Images